Project

General

Profile

1 44133 michele.ar
package eu.dnetlib.data.mdstore.plugins;
2
3 49368 michele.ar
import java.io.StringReader;
4 49210 michele.ar
import java.net.URI;
5
import java.net.URISyntaxException;
6 55934 michele.ar
import java.util.ArrayList;
7
import java.util.Arrays;
8 55101 michele.ar
import java.util.HashMap;
9 49368 michele.ar
import java.util.List;
10 51852 michele.ar
import java.util.Map;
11 49380 michele.ar
import java.util.Set;
12 51020 michele.ar
import java.util.stream.Collectors;
13 44133 michele.ar
14 59332 michele.ar
import org.apache.commons.lang3.StringUtils;
15 49368 michele.ar
import org.apache.commons.logging.Log;
16
import org.apache.commons.logging.LogFactory;
17 49210 michele.ar
import org.dom4j.Document;
18 49368 michele.ar
import org.dom4j.DocumentException;
19 49380 michele.ar
import org.dom4j.Element;
20 49368 michele.ar
import org.dom4j.Node;
21
import org.dom4j.io.SAXReader;
22 60864 michele.ar
import org.springframework.beans.factory.annotation.Autowired;
23 49368 michele.ar
import org.springframework.beans.factory.annotation.Value;
24 44133 michele.ar
25 55934 michele.ar
import com.google.common.base.Splitter;
26 60864 michele.ar
import com.mongodb.MongoClient;
27
import com.mongodb.client.MongoCollection;
28 55934 michele.ar
29 51020 michele.ar
import eu.dnetlib.data.mdstore.plugins.objects.MdRecord;
30 49416 michele.ar
import eu.dnetlib.data.mdstore.plugins.objects.MyURL;
31 51020 michele.ar
import eu.dnetlib.data.mdstore.plugins.objects.Project;
32 61487 michele.ar
import eu.dnetlib.data.utils.XsltFunctions;
33 49416 michele.ar
34 49291 michele.ar
public class EnrichOpenairePlugin extends GenericDoiMdstorePlugin {
35 49210 michele.ar
36 49368 michele.ar
	private static final Log log = LogFactory.getLog(EnrichOpenairePlugin.class);
37
38
	@Value("${plugin.enrich.publications.openaire.url}")
39
	private String baseUrl;
40
41 60102 michele.ar
	@Value("${plugin.enrich.openaire.datasources.blacklist}")
42
	private String datasourceBlackList;
43
44 60864 michele.ar
	@Autowired
45
	private MongoClient mongoClient;
46
47 55101 michele.ar
	private Map<String, Counter> counters = new HashMap<>();
48
49 49291 michele.ar
	@Override
50
	protected URI prepareURI(final String doi) throws URISyntaxException {
51 49368 michele.ar
		return new URI(String.format(baseUrl, doi));
52 49291 michele.ar
	}
53 44133 michele.ar
54
	@Override
55 60864 michele.ar
	protected MongoCollection<org.bson.Document> getCacheCollection() {
56
		return mongoClient.getDatabase("API_CACHES").getCollection("OPENAIRE_API_CACHE");
57
	}
58
59
	@Override
60 55101 michele.ar
	protected void reconfigure(final Map<String, String> params) {
61
		counters.clear();
62
		counters.put("subjects", new Counter());
63
		counters.put("citations", new Counter());
64
		counters.put("urls", new Counter());
65
		counters.put("projects", new Counter());
66 59928 michele.ar
		counters.put("dois", new Counter());
67 55101 michele.ar
	}
68 44133 michele.ar
69 51020 michele.ar
	@Override
70 55101 michele.ar
	protected void resetConfiguration() {
71
		log.info("***** Openaire Enrichment - subjects  : " + counters.get("subjects"));
72
		log.info("***** Openaire Enrichment - citations : " + counters.get("citations"));
73
		log.info("***** Openaire Enrichment - urls      : " + counters.get("urls"));
74
		log.info("***** Openaire Enrichment - projects  : " + counters.get("projects"));
75 59928 michele.ar
		log.info("***** Openaire Enrichment - dois      : " + counters.get("dois"));
76 55101 michele.ar
		counters.clear();
77
	}
78 54968 michele.ar
79
	@Override
80 51020 michele.ar
	protected boolean updateDocument(final MdRecord doc, final String response) {
81 55101 michele.ar
		counters.get("subjects").incrementBefore(doc.getSubjects().size());
82
		counters.get("citations").incrementBefore(doc.getCitations().size());
83
		counters.get("urls").incrementBefore(doc.getUrls().size());
84
		counters.get("projects").incrementBefore(doc.getProjects().size());
85 59928 michele.ar
		counters.get("dois").incrementBefore(doc.getDois().size());
86 51020 michele.ar
87 49368 michele.ar
		try {
88
			final Document docRes = (new SAXReader()).read(new StringReader(response));
89 49210 michele.ar
90 49368 michele.ar
			final List<?> results = docRes.selectNodes("/response/results/result");
91 44133 michele.ar
92 49368 michele.ar
			if (results.size() == 1) {
93
				final Node n = (Node) results.get(0);
94 49380 michele.ar
				updateSubjects(doc, n);
95
				updateCitations(doc, n);
96
				updateUrls(doc, n);
97 49395 michele.ar
				updateProjects(doc, n);
98 59928 michele.ar
				updateDois(doc, n);
99 51020 michele.ar
				updateBestRights(doc);
100 55101 michele.ar
101 49368 michele.ar
				return true;
102
			} else if (results.size() == 1) {
103
				log.warn("Too many responses");
104
			}
105
		} catch (final DocumentException e) {
106
			log.warn("Invalid response", e);
107 55101 michele.ar
		} finally {
108
			counters.get("subjects").incrementAfter(doc.getSubjects().size());
109
			counters.get("citations").incrementAfter(doc.getCitations().size());
110
			counters.get("urls").incrementAfter(doc.getUrls().size());
111
			counters.get("projects").incrementAfter(doc.getProjects().size());
112 59928 michele.ar
			counters.get("dois").incrementAfter(doc.getDois().size());
113 49368 michele.ar
		}
114
115
		return false;
116 49210 michele.ar
	}
117 49380 michele.ar
118 51020 michele.ar
	private void updateSubjects(final MdRecord doc, final Node n) {
119 55934 michele.ar
		final Set<String> subjects = doc.getSubjects()
120
				.stream()
121
				.map(EnrichOpenairePlugin::cleanSubject)
122
				.flatMap(List::stream)
123
				.collect(Collectors.toSet());
124
125 49380 michele.ar
		for (final Object o : n.selectNodes(".//subject[@classid='keyword']")) {
126 55934 michele.ar
			subjects.addAll(cleanSubject(((Node) o).getText().trim()));
127 49380 michele.ar
		}
128 55934 michele.ar
129 51020 michele.ar
		doc.setSubjects(subjects);
130 49380 michele.ar
	}
131
132 55934 michele.ar
	public static List<String> cleanSubject(final String s) {
133
		if (s.isEmpty()) {
134
			return new ArrayList<>();
135
		} else if (s.startsWith("info:eu-repo/classification/msc/")) {
136
			return new ArrayList<>();
137
		} else if (s.startsWith("info:eu-repo/classification/acm/")) {
138
			return Arrays.asList(s.replaceFirst("info:eu-repo/classification/acm/", ""));
139
		} else if (s.contains(";")) {
140
			return Splitter.on(";").trimResults().omitEmptyStrings().splitToList(s);
141
		} else if (s.contains(",")) {
142
			return Splitter.on(",").trimResults().omitEmptyStrings().splitToList(s);
143
		} else {
144
			return Arrays.asList(s);
145
		}
146 49804 michele.ar
	}
147
148 51020 michele.ar
	private void updateCitations(final MdRecord doc, final Node n) {
149
		doc.getCitations().clear();
150 49380 michele.ar
151
		for (final Object o : n.selectNodes(".//citations/citation/rawText")) {
152 51020 michele.ar
			doc.getCitations().add(((Node) o).getText());
153 49380 michele.ar
		}
154 62539 michele.ar
		for (final Object o : n.selectNodes(".//references/reference/rawText")) {
155
			doc.getCitations().add(((Node) o).getText());
156
		}
157 49380 michele.ar
	}
158
159 51020 michele.ar
	private void updateUrls(final MdRecord doc, final Node n) {
160
		doc.getUrls().addAll(doc.getUrls());
161 49380 michele.ar
162 60102 michele.ar
		final Set<String> blacklist =
163
				Arrays.stream(datasourceBlackList.split(",")).map(String::trim).filter(StringUtils::isNotBlank).collect(Collectors.toSet());
164
165 49380 michele.ar
		for (final Object oin : n.selectNodes(".//instance")) {
166 60102 michele.ar
167
			final String hostedByid = ((Element) oin).valueOf("./hostedby/@id").trim();
168
169
			if (!blacklist.contains(hostedByid)) {
170 61593 michele.ar
				final String hostedBy = ((Element) oin).valueOf("./hostedby/@name").trim();
171 60102 michele.ar
				final String rights = ((Element) oin).valueOf("./accessright/@classname").trim();
172
173
				for (final Object ourl : ((Element) oin).selectNodes("./webresource/url")) {
174 61593 michele.ar
					final String url = ((Node) ourl).getText().trim();
175
					final String name =
176
							hostedBy.equalsIgnoreCase("Unknown") || hostedBy.equalsIgnoreCase("Unknown Repository") ? XsltFunctions.serverName(url) : hostedBy;
177
					final MyURL u = new MyURL(url, name, rights);
178 60102 michele.ar
					doc.getUrls().remove(u);
179
					doc.getUrls().add(u);
180
				}
181 49380 michele.ar
			}
182
		}
183
	}
184 49395 michele.ar
185 51020 michele.ar
	private void updateProjects(final MdRecord doc, final Node n) {
186 49416 michele.ar
187 49395 michele.ar
		for (final Object op : n.selectNodes(".//rels/rel[./to/@type='project']")) {
188
			final Node p = (Node) op;
189 59332 michele.ar
			final String name = p.valueOf("./title").trim();
190 49395 michele.ar
191 59332 michele.ar
			if (StringUtils.isNotBlank(name) && !name.equalsIgnoreCase("null") && !name.equalsIgnoreCase("unidentified") && !name.equalsIgnoreCase("unknown")) {
192
				final Project np = new Project();
193
				np.setOpenaireId(p.valueOf("./to"));
194
				np.setCode(p.valueOf("./code"));
195
				np.setName(name);
196
				np.setAcronym(p.valueOf("./acronym"));
197
				np.setFunder(p.valueOf(".//funder/@shortname"));
198
				np.setProgram(p.valueOf(".//funding_level_0/@name"));
199
				np.setJurisdiction(p.valueOf(".//funder/@jurisdiction"));
200 61487 michele.ar
				np.setInfoId(XsltFunctions.projectLongId(np.getFunder(), np.getProgram(), np.getCode(), np.getJurisdiction(),
201 59332 michele.ar
						np.getName(), np.getAcronym()));
202
203
				doc.getProjects().add(np);
204
			}
205 49395 michele.ar
		}
206 50059 michele.ar
	}
207 49395 michele.ar
208 59928 michele.ar
	private void updateDois(final MdRecord doc, final Node n) {
209 59937 michele.ar
		for (final Object od : n.selectNodes(".//*[local-name()='result']/pid[@classid='doi']")) {
210 61531 michele.ar
			final String doi = XsltFunctions.cleanDoi(((Node) od).getText());
211 59928 michele.ar
			if (StringUtils.isNotBlank(doi)) {
212
				doc.getDois().add(doi);
213
			}
214
		}
215
	}
216
217 51020 michele.ar
	private void updateBestRights(final MdRecord doc) {
218
		final Set<String> availables = doc.getUrls().stream().map(MyURL::getRights).map(String::toUpperCase).collect(Collectors.toSet());
219 50059 michele.ar
		if (availables.contains("OPEN ACCESS")) {
220 51020 michele.ar
			doc.setBestRights("Open Access");
221 50059 michele.ar
		} else if (availables.contains("EMBARGO")) {
222 51020 michele.ar
			doc.setBestRights("Embargo");
223 50059 michele.ar
		} else if (availables.contains("RESTRICTED")) {
224 51020 michele.ar
			doc.setBestRights("Restricted");
225 50059 michele.ar
		} else if (availables.contains("CLOSED ACCESS")) {
226 51020 michele.ar
			doc.setBestRights("Closed Access");
227 50059 michele.ar
		} else {
228 51020 michele.ar
			doc.setBestRights("Unknown");
229 50059 michele.ar
		}
230 49395 michele.ar
	}
231
232 44133 michele.ar
}